lewtun's picture
lewtun HF staff
Add evaluation results on the autoevaluate--squad-sample config of autoevaluate/squad-sample
d7cbdd9
|
raw
history blame
1.39 kB
metadata
language: en
datasets:
  - squad
metrics:
  - squad
license: apache-2.0
model-index:
  - name: autoevaluate/distilbert-base-cased-distilled-squad
    results:
      - task:
          type: question-answering
          name: Question Answering
        dataset:
          name: autoevaluate/squad-sample
          type: autoevaluate/squad-sample
          config: autoevaluate--squad-sample
          split: test
        metrics:
          - name: Exact Match
            type: exact_match
            value: 84
            verified: true
          - name: F1
            type: f1
            value: 87.8248
            verified: true

DistilBERT base cased distilled SQuAD

Note: This model is a clone of distilbert-base-cased-distilled-squad for internal testing.

This model is a fine-tune checkpoint of DistilBERT-base-cased, fine-tuned using (a second step of) knowledge distillation on SQuAD v1.1. This model reaches a F1 score of 87.1 on the dev set (for comparison, BERT bert-base-cased version reaches a F1 score of 88.7).

Using the question answering Evaluator from evaluate gives:

{'exact_match': 79.54588457899716,
 'f1': 86.81181300991533,
 'latency_in_seconds': 0.008683730778997168,
 'samples_per_second': 115.15787689073015,
 'total_time_in_seconds': 91.78703433400005}

which is roughly consistent with the official score.